thicc_girls_are_teh_best
Member
Btw, in Oberon regression tests they also cut of ROPs to 32, as well as bandwidth to 218GB/s in GEN1 (and 176GB/s in GEN0 mode).
On top of that, in Oberon_Native test its very clear that full chip is 36CUs, as can be found in comment section beside a test concerning BW when 18WGPs (36CUs) are used. Too many things are fitting for people to blindly hand waive it not the least the fact that all data was deleted from Git and Twitter.
The main reason I've seen people dismissing those tests is because of lack of ray-tracing (and unfortunately that caused some to try speculating PS5 didn't even have ray-tracing, which is a ridiculous assertion). But at the same time, why would they need ray-tracing enabled in the first place for PS4 and PS4 Pro back-compat testing?
They wouldn't, so it wasn't listed. Pretty simple conclusion. Now if there's some other PS5 chip out there that's had its benchmark testing kept more concealed than confidential intel at the NSA, and we happen to get some results testing of that leaked out sometime soon, then I'll gladly buy into that as very possible PS5 performance metrics.
That door is closing kinda fast, though. Hopefully we get some solid info from both Sony and MS on this stuff relatively soon (beyond what's already been provided, anyway).